World Development Vol. 32, No. 1, pp.

137157, 2004
2003 Elsevier Ltd. All rights reserved
Printed in Great Britain
0305-750X/$ - see front matter

www.elsevier.com/locate/worlddev

doi:10.1016/j.worlddev.2003.07.005

Agricultural Expansion, Resource Booms and

Growth in Latin America: Implications for
Long-run Economic Development
EDWARD B. BARBIER *
University of Wyoming, Laramie, WY, USA
Summary. This paper examines a number of key issues concerning agricultural land expansion,
resource booms and economic growth in Latin America. The structural characteristics of
agricultural development, represented by cropland share of total land area, agricultural export
share of total exports, and cereal yields, appear to inuence agricultural land expansion in the
region. Long-run agricultural land expansion may also be correlated with a boom and bust
pattern of economic development. Contributing factors include distorted land and resource
markets, ineective property rights, and the failure to target policies to improve the ecient and
sustainable management of natural resources.
2003 Elsevier Ltd. All rights reserved.
Key words agricultural land expansion, Dutch disease, economic growth, Latin America,
Resource booms

1. INTRODUCTION
Recent research suggests three worrying
trends with regard to the role of natural
resources in long-term economic development
in Latin America. First, natural-resource commodity booms in the region appear to be
accompanied by declining GDP per capita.
This may be the result of a potential Dutch
disease eect; i.e., the impact of the boom is
to divert economic resources from more innovative, manufacturing sectors to less innovative, primary sectors. Second, rapid land use
changes are occurring in Latin America, characterized mainly by the conversion of tropical
forests and other natural habitat to agriculture.
Empirical evidence suggests however, that over
the long run agricultural land expansion may
be correlated with lower, rather than higher,
levels of GDP per capita. Finally, excessive
exploitation of land and natural resources
across the region is being promoted by policies
that distort land and resource markets, ineffective property rights and the failure to target
policies to improve the ecient and sustainable
management of natural resources.
This paper explores the economic determinants of these three key trends concerning
agricultural land expansion, resource booms
137

and growth in Latin America. Specically, the

paper focuses on recent empirical and theoretical work that addresses the following questions:
What are the main factors determining
long-run agricultural land expansion in
Latin America compared to other tropical
regions?
Given the importance of natural resourcebased sectors for most economies in Latin
America, what has been the impact of

* Earlier

versions of this paper were presented at The

World Bank Session on Measurement Issues in Natural
Resources, Development and Poverty at the Conference on Natural Capital, Poverty and Development.
The Munk Centre for International Studies, University
of Toronto, Canada, September 58, 2001 and at the
Workshop on Growth and Environment: Macro-Economic Perspectives, Inter-American Development
Bank, Washington, DC, May 29, 2001. I am grateful for
comments provided by Walter Arensberg, Kirk Hamilton, Glenn-Marie Lange, Ramn Lpez, Diego Rodriguez
and an anonymous referee, as well as to participants at
the above two meetings. The paper beneted from
research assistance provided by Seong-Hee Kim and
editorial assistance by Margie Reis. Final revision
accepted: 23 July 2003.

138

WORLD DEVELOPMENT

price-induced resource booms on economic growth in the region?

Is there also an inherent boom and bust
pattern of economic development associated
with agricultural land expansion, and if so,
do economic policies in Latin America exacerbate this problem?
The outline of the paper is as follows. Section
2 provides a brief overview of land use and
deforestation trends in all tropical regions as
well as in Latin America. Section 3 discusses a
crosscountry analysis of the eects of growth,
income per capita and other macroeconomic
factors on agricultural land expansion in
developing regions, including Latin America.
Section 4 summarizes recent theories and evidence as to why resource booms in Latin
America may not lead to higher growth rates,
and examines the relationship between longrun agricultural land expansion in Latin
America and economic development. Section 5
looks at the role of current economic policies in
inuencing these relationships. The concluding
section discusses the possible policy implications for Latin America.

2. LAND USE AND DEFORESTATION

TRENDS IN ALL TROPICAL COUNTRIES
AND LATIN AMERICA
Table 1 displays global tropical deforestation
trends over 19802000. There appears to be a
substantial dierence in deforestation, particu-

larly in Latin America, between the 198190

and 19912000 periods. Over 198190, the area
of tropical forests cleared on average each year
in Latin America was 7.4 million ha, almost as
much as the area deforested in Asia and Africa
put together. Within Latin America, the main
source of total deforestation occurred in South
America, around 6.2 million ha annually. But,
the fastest rate of deforestation was taking
place in Central America and Mexico, with
1.5% of the forests in the region being cleared
annually. Over 19912000, tropical deforestation appears to have slowed globally, and
especially in Latin America. Deforestation in
the region has declined by over 40% to 4.3
million ha annually. The sharpest decline has
occurred in South America, where annual
deforestation has fallen to 3.4 million ha, or
0.4% per year. Central America and Mexico
still has the fastest rate of deforestation, at
1.4% per year, and approximately 1 million ha
is being cleared annually in this region (mainly
in Mexico). In the Caribbean, the modest
deforestation across the region appears to have
been oset by an increase in forest area in Cuba
by 28,000 ha.
Total deforestation across Latin America
remains concentrated in a few countries: Brazil
(2.2 million ha annually, or a 0.4% deforestation rate), Mexico (0.6 million ha, or a 1.1%
rate), Peru (0.3 million ha, or a 0.4% rate),
Colombia (0.2 million ha, or a 0.4% rate), and
Bolivia (0.2 million ha, or a 0.3% rate). This is
not surprising, given that these ve countries
contain 86% of the total tropical forest area

Table 1. Global tropical deforestation trends, 19802000

Region

Africa
Asia and Pacic
Latin America and
Caribbean
C. America and
Mexico
Caribbean
Trop. South
America
Total

Land area
(million
ha)

Forest cover

Annual
deforestation
198190

Estimated
annual deforestation
199120

1980
(million
ha)

1990
(million
ha)

Million ha

% per
annum

Million ha

% per
annum

2,236.1
892.1
1,650.1

568.6
349.6
992.2

527.6
310.6
918.1

4.1
3.9
7.4

0.7
1.2
0.8

5.3
2.5
4.3

1.0
0.8
0.5

239.6

79.2

68.1

1.1

1.5

0.97

1.4

69.0
1,341.6

48.3
864.6

47.1
802.9

0.1
6.2

0.3
0.7

)0.02
3.37

)0.1
0.4

4,778.3

1,910.4

1,756.3

15.4

0.8

12.2

0.7

Source: FAO (1993), except for estimated annual deforestation 19912000 (FAO, 2001).

AGRICULTURAL EXPANSION AND RESOURCE BOOMS

found in Latin America. In fact, Brazil alone

accounts for 60% of Latin Americas tropical
forests. Thus the slowdown in deforestation in
Brazil over 19912000 is largely responsible for
the decline in overall tropical deforestation in
Latin America compared to the 198190
period.
The available evidence suggests that in most
developing economies the decline in forest and
woodlands is mainly the result of land conversion, in particular agricultural expansion for
crop production (FAO, 1997). Land expansion
occurring in tropical regions appears to be
related to structural features of the agricultural
sectors of developing economies, such as low
irrigation and fertilizer use as well as poor crop
yields. Low agricultural productivity and input
use reect poor agricultural intensication and
development, which in turn mean more pressure is put on conversion of forests and other
marginal lands for use in crop production.
Table 2 indicates both the current and future
dependence of developing countries on agricultural land expansion for crop production.
Over 197090 increased harvested area
accounted for 48% of the additional crop production in Latin America. This dependence on
expanding harvested land was the highest for
any developing region over this period.
Although improved cropping intensity and
yields are expected to reduce Latin Americas
dependency on agricultural land expansion
over 19902010, about 29% of the contribution
to total crop production increases in the region
will be derived from expansion of cultivated
land.
Fischer and Heilig (1997) combined the
results of the FAO (1995) study summarized in
Table 2 with recent UN population projections
to estimate the demand for additional cultivated land in developing countries in 2050.
Their results are indicated in Table 3. All
developing countries are expected to increase
their demand for cultivated cropland considerably, leading to extensive conversion of forests and wetlands. In South America, cultivated
land area is expected to increase by over 50%
by 2050, with about 70% of the new land
coming from deforestation and wetland conversion. Agricultural land will expand by
around 40% in Central America and the
Caribbean, but 80% of this new land is likely to
be from forest and wetland conversion.
Pressures on the land and natural resource
base in developing countries are also likely to
come from the overall demands of economic

139

development. For example, many low-income

and lower middle-income economies are highly
resource dependent (Barbier, 1999). Not only do
these economies rely principally on direct
exploitation of their resource bases through
primary industries (e.g., agriculture, forestry,
shing, etc.) but also over 50% or more of their
export earnings come from a few primary
commodities. Natural capitalthe value of the
natural resource endowment of a countryis
particularly important in the developing world.
For low-income countries dependent on export
revenues from primary commodities (other
than petroleum), 20% of their national wealth
comprises natural capital (World Bank,
1997). 1 In the poorest countries, agricultural
cropland comprises around 80% of the natural
capital wealth. The share of natural capital in
total wealth is 11% in the Caribbean, 9% in
South America and 6% in Central America.
Clearly, exploitation of the natural resource
and land base of these countries is essential for
their long-term economic development eorts.
3. FACTORS DETERMINING
AGRICULTURAL LAND EXPANSION
The discussion of the previous section suggests that the major cause of forest loss in
developing countries is conversion to agriculture. Thus a crosscountry analysis of agricultural land expansion should also provide
insights into the factors inuencing tropical
deforestation. Equally, previous studies of
tropical deforestation may be able to suggest
some of the possible eects of growth, income
per capita and other macroeconomic factors on
agricultural land expansion in developing
regions, including Latin America. Four distinct
analytical frameworks have been proposed in
the economics literature for motivating crosscountry estimations of the causes of agricultural
land conversion and tropical deforestation: the
environmental Kuznets curve hypothesis,
competing land use models, forest land conversion
models, and institutional models (Barbier &
Burgess, 2001). As the following brief review
indicates, these analytical frameworks enable us
to focus on certain key economic factors that
may determine tropical agricultural land
expansion and to choose the appropriate variables to include in our crosscountry regression. 2
The environmental Kuznets curve (EKC)
hypothesis states that an environmental bad

140

Table 2. Trends in crop production and harvest area in developing regions

Region

Crop production

Sub-Saharan Africa
Near East and North Africa
East Asiaa
South Asia
Latin Americab
All developing countries
Source: FAO (1995).
a
Excludes China.
b
Includes the Caribbean.

Harvested land

19902010 Contribution (%) of

increases in

19902010 Contribution (%) of

increases in

Yields

Harvested area

Yields

Harvested area

Arable land

Cropping intensity

53
73
59
82
52
69

47
27
41
18
48
31

53
71
61
82
53
66

47
29
39
18
47
34

64
31
82
22
60
62

36
69
18
78
40
38

19902010
Percentage of
crop production
from new land

30
8
34
4
29
19

WORLD DEVELOPMENT

197090 Contribution (%) of

increases in

AGRICULTURAL EXPANSION AND RESOURCE BOOMS

141

Table 3. Demand for cultivated land in 2050 in developing regions

Region

Cultivated crop
land in 1990
(1000 ha)

% of production
increase from new
land

Additional cultivated
land required in 2050
(1000 ha)

% of new lands
from forest and
wetland conversion

Africa
Asiaa
Latin America
Central Americab
South America

252,583
456,225
189,885
36,920
152,965

29
10
28
20
30

241,703
85,782
96,710
14,603
82,107

61
73
70
80
69

All developing
countries

899,795

21

424,194

66

Source: Fischer and Heilig (1997).

a
Excludes China.
b
Includes the Caribbean.

rst increases, but eventually falls, as the per

capita income of a country rises. There are a
number of recent theoretical models explaining
why such an inverted-U relationship between
income and environmental bads might hold
(e.g., Andreoni & Levinson, 2001; McConnell,
1997; Stokey, 1998). Although the EKC model
has generally been applied to pollution problems, there have been a number of recent
studies that have also examined whether this
hypothesis also holds for global deforestation
(e.g., Antle & Heidebrink, 1995; Cropper &
Griths, 1994; Koop & Toole, 1999; Panayotou, 1995; Shak, 1994). The basic EKC model
for deforestation is usually
Fit
Fit
1 F Yit ; Yit2 ; zit
a1 Yit
a2 Yit2 zit b eit ;

where Fit
Fit
1 is the change in the forest stock

over the previous period (which is negative if
deforestation is occurring), Yit is per capita
income and zit is a 1  n vector that includes
other explanatory variables, such as population
density or growth and other macroeconomic
variables. 3
The application of the EKC model (1) to
explain deforestation trends across countries
has produced mixed results. When the model is
tested for both temperate and tropical countries, it is inconclusive (Antle & Heidebrink,
1995; Panayotou, 1995; Shak, 1994). When
applied to just tropical countries, the invertedU relationship tends not to hold for all countries but may apply to specic regions. For
example, Cropper and Griths (1994) nd
some evidence that the EKC model is relevant
to Latin America and Africa. But, for each of
these regions the turning pointthe per capita

income level at which the deforestation rate is

zero and is about to declineis generally two
to four times higher than the average per capita
income for that region. Nevertheless, as Table 1
has indicated, tropical deforestation in most
developing regions, especially Latin America,
has slowed down recently, and one possible
explanation may be the EKC hypothesis.
Other empirical analyses have taken as their
starting point the hypothesis that forest loss in
tropical countries is the result of competing
land use, in particular between maintaining the
natural forest and agriculture (e.g., Barbier &
Burgess, 1997; Ehui & Hertel, 1989). As indicated in Tables 2 and 3, the evidence across
tropical regionsand in particular Latin
Americais that substantial conversion of
forest and woodlands to agriculture is occurring. From an economic standpoint, given the
time and eort required to reestablish tropical
forest (where this is ecologically feasible) such
conversion implies that potential timber and
environmental benets from forestland are
irreversibly lost. Therefore, competing land use
models usually include some measure of the
price or opportunity cost of agricultural
conversion and deforestation in terms of the
foregone benets of timber production and
environmental benets from forestland
Fit
Fit
1 AD vit ; zit ;

oAD =ovit < 0;

where vit is the opportunity cost or price of

agricultural conversion, AD is the demand for
converting forest land to agriculture, and as
before zit is a vector containing exogenous
economic factors (e.g., income per capita,
population density, agricultural yields).
Many country-level studies of tropical
deforestation have focused on modeling the

142

WORLD DEVELOPMENT

forestland conversion decision of agricultural

households. There have been several such
applications to Latin American countries (e.g.,
Anderson, Granger, Eustaquio, Weihold, &
Wunder, 2002; Barbier, 2002; Barbier & Burgess, 1996; Chomitz & Gray, 1996; Nelson,
Harris, & Stone, 2001). Such approaches model
the derived demand for converted land by rural
smallholders, and assume that the households
either use available labor to convert their own
land or purchase it from a market. This in turn
allows the determinants of the equilibrium level
of converted land to be specied. In such
models, the aggregate equilibrium level of
cleared land across all households is usually
hypothesized to be a function of output and
input prices and other factors aecting aggregate conversion
D
AD
it A pit ; wLit ; wit ; xit ; zit ;

oAD
< 0;
owLit

oAD
> 0;
oxit

oAD
> 0;
opit
3

where p is the price of agricultural output, wL is

rural wage (labor is a key component in land
clearing), w is a vector of other inputs, x are
factors inuencing the accessibility of forest
areas (e.g., roads, infrastructure, distance to
major towns and cities), and as before zit represents other economic explanatory variables.
Although both the competing land use and
forestland conversion models appear to work
well for specic tropical forest countries, it is
dicult to obtain time series data across
countries for key price data in the respective
models, i.e., vit in Eqn. (2) or pit and especially
wit in Eqn. (3). Crosscountry data on important
x variables, such as rural road expansion and
road building investments, are also hard to
nd. As a result, crosscountry analyses of Eqns.
(2) and (3) have tended to leave out prices and
x factors, or employed proxies. For example,
in their empirical estimations for deforestation
across all tropical countries, Barbier and Burgess (1997) employed roundwood production
per capita as a proxy for vit in Eqn. (2), as
preferred measures of the opportunity cost
of conversion (e.g., land values, timber rents)
are not available across countries. Similarly,
Southgate (1994) used annual population
growth, agricultural export growth, crop yield
growth and a land constraint dummy to explain
annual agricultural land growth across Latin
America over 198287. He found that population and agricultural export growth were posi-

tively related to land expansion, whereas yield

growth and the land constraint were negatively
related. Other studies have also demonstrated
that structural agricultural, economic and
geographic factors, which vary from country to
country, are signicant in explaining the different land conversion trends across countries
(e.g., Barbier & Burgess, 1997; Deacon, 1999).
These structural variables include agricultural yield, cropland share of land area, agricultural export share, and arable land per
capita, which capture country-by-country differences in agricultural sectors and land use
patterns, as well as other exogenous macroeconomic variables, such as population growth,
rural population density, GDP growth, real
interest rates and debt. These factors may be
particularly signicant explanatory variables in
a crosscountry analysis, if the diculty in
obtaining crosscountry time series data on key
variables, such as rural wages, roads, other
input prices, makes it impossible to include
variables representing agricultural returns or
accessibility of forest lands in the model.
Finally, recent empirical analyses at both the
country and crosscountry level have explored
the impact on tropical deforestation of institutional factors, such as land use conict, security
of ownership or property rights, political stability, and the rule of law (e.g., Alston, Libecap, & Mueller, 1999, 2000; Deacon, 1994,
1999; Godoy et al., 1998). The main hypothesis
tested is that such institutional factors are
important factors explaining deforestation
Fit
Fit
1 F qit ; zit ;

where qit is a vector of institutional factors and

and zit represents other economic explanatory
variables.
Although such models have demonstrated
the importance of institutional factors in
determining deforestation, it is unclear how
much weight should be given to such factors in
preference to explanatory variables identied
by other approaches to crosscountry analyses
of forest loss. Nevertheless, the failure to
include institutional factors in a crosscountry
analysis of land use change may mean that
potentially important explanatory variables
have been omitted.
In sum, the four main analytical frameworks
motivating crosscountry analysis of tropical
deforestation and land use change emphasize
the following key variables: (a) from Eqn. (1),
the inclusion of per capita income and income
squared terms to test for a possible EKC rela-

AGRICULTURAL EXPANSION AND RESOURCE BOOMS

tionship; (b) in the absence of adequate crosscountry data for the price and x variables in
Eqns. (2) and (3), the inclusion of the synthesis
model should include certain structural
variables sit , such as agricultural yield, cropland share of land area, agricultural export
share, and arable land per capita, to capture
country-by-country dierences in agricultural
sectors and land use patterns; and nally, (c)
from Eqn. (4) the inclusion of key institutional
factors thought to inuence land expansion and
deforestation.
Thus a possible model for a crosscountry
analysis of the possible eects of growth,
income per capita and other macroeconomic
factors on agricultural land expansion in
developing regions might look like:
Ait
Ait
1 AYit ; Yit2 ; sit ; zit ; qi ;
for country i at time t;

where Ait
Ait
1 represents expansion in agricultural land area, Yit is per capita income, sit is
a vector of structural variables, such as
agricultural yield, cropland share of land area,
agricultural export share, and arable land per
capita, and zit represents other exogenous
explanatory variables, such as population density or growth, GDP growth and other macroeconomic variables. Finally, as institutional
factors qi tend to be invariant with time, two
versions of the model can be tested, one without and one including qi .
Eqn. (5) was estimated through a panel
analysis of tropical agricultural land expansion
in Latin America over 196194, with the
dependent variable being the percentage annual
change in agricultural land area. 4 The EKC
variables Yit ; Yit2 are represented by gross
domestic product (GDP) per capita in constant
purchasing power parity (1987 $) and by GDP
per capita squared, respectively. The structural
variables sit are cereal yield, cropland share of
total land area, agricultural export share of
total merchandise exports and arable land per
capita. The additional explanatory variables
zit are population and GDP growth. The
source of data used for these variables was the
World Banks World Development Indicators,
which has the most extensive data set for key
land, agricultural and economic variables for
developing countries over the period of analysis.
Table 4 indicates the results for Latin
America both without institutional factors, qi ,
and with the inclusion of three institutional

143

variables: a corruption index, a property rights

index and a political stability index. These
indices were obtained from the Levine
LoayzaBeck data set used in Beck, Levine, and
Loayza (2000) and Levine, Loayza, and Beck
(2000), which are available from the Economic
Growth Research Group of the World Bank.
The corruption and property rights indices are
directly from the LevineLoayzaBeck data set
and are averaged over 198295. The political
stability index was created as a composite index
of the average number of revolutions and coups
(averaged over 196090), average number of
assassinations per million population (averaged
over 196090), and an index of ethnic fractionalization (averaged over 198295). As these
indices were not available for all tropical
countries in the original sample, the inclusion
of these institutional factors reduced the sample
sizes of the regressions considerably. In addition, the three indices are time invariant, and
with their inclusion in addition to the original
explanatory variables of the model, xed eects
regressions cannot be run. 5
Both one-way and two-way xed and random
eects models were tested for the sample of
Latin America countries. Table 4 displays the
results for the preferred models and the relevant
statistics. In the table, the parameter estimates
are reported as elasticities, in order to facilitate
comparison of the eects of the dierent variables, which are in dierent units. In the
regression without institutional indices, only the
structural variables appear to explain agricultural expansion in Latin America. Growth in
agricultural land area increases with cropland
share of total land area and agricultural export
share but declines with cereal yield. But, neither
per capita income nor GDP per capita squared
is signicant. Thus, the results of this regression
lead to rejection of the EKC hypothesis for
explaining the pattern of agricultural land
expansion in Latin America.
In comparison, the inclusion of institutional
factors has a considerable inuence on the
analysis. First, the EKC hypothesis can no
longer be rejected, although the estimated EKC
turning point of $4,946 is one third larger than
the average per capita income of $3,675 for the
sample of Latin American countries. Thus
there is some evidence for Latin America that
agricultural land expansion will start to slow
down as per capita income rises in the region,
but clearly most countries are still well below
the turning point level of income for this to
occur. Second, of the institutional variables,

144

WORLD DEVELOPMENT
Table 4. Panel analysis of agricultural land expansion in tropical Latin America, 196194
Dependent variable: agricultural land expansion
(% annual changea )
Elasticity estimatesb
Without institutional factors

With institutional factors

(N 319)
)0.388
()0.109)
0.050
(0.036)
0.009
(0.127)
)0.734
()1.234)
)2.632
()2.028)*
5.565
(2.485)**
0.462
(2.098)*
)0.263
()0.163)

(N 233)
7.943
(2.595)**
)3.499
()2.328)*
)0.066
()0.862)
1.421
(2.271)*
)0.623
()0.929)
0.789
(2.343)*
0.383
(2.948)**
0.358
(0.83)
0.249
(0.345)
)0.363
()0.425)
0.494
(2.220)*

No
($17,359)
3.323**
11.23**
15.52*

Yes
($4,946)

Explanatory variables
GDP per capitac
(PPP, constant 1987 $)
GDP per capita squared
GDP growth
(% annual change)
Population growth
(% annual change)
Cereal yield
(kg per hectare)
Cropland share of land
(% of land area)
Agricultural export share
(% of merchandise exports)
Arable land per capita
(hectares per person)
Corruption index
(high 0, low 10)
Property rights index
(high 5, low 1)
Political stability index
(high 0, low 1)
Kuznets curve
(Turning point estimate)
F -test for pooled model
BreuschPagan (LM) test
Hausman test
DurbinWatson statistic
Adjusted R2
Preferred model

2.045
0.128
One-way xed eects

OLS

Mean for the sample without institutional factors is 0.75%; mean for the sample with institutional factors is 0.66%.
t-Ratios are indicated in parentheses.
c
Mean for the sample without institutional factors is $3,654; mean for the sample with institutional factors is $3,675.
PPP is purchase power parity.
*Signicant at 5% level.
**Signicant at 1% level.
b

political instability appears to have a signicant

and positive impact on agricultural expansion
in Latin America. 6 Third, some structural
variables are still important in explaining land
expansion. Growth in agricultural land area in
Latin America increases with cropland share of
total land area and agricultural export share.
Finally, population growth appears also to be
associated with greater land expansion in the
region.

The above results for the synthesis model

provide interesting additional insights into
agricultural land expansion in Latin America.
First, the pattern of agricultural development,
as represented by such structural variables as
cropland share of total land area, agricultural
export share of total exports, and to some
extent, cereal yields, appears consistently to
inuence agricultural land expansion. This
would suggest that policies that inuence the

AGRICULTURAL EXPANSION AND RESOURCE BOOMS

structure of agricultural production and overall

development in the region could have a
substantial inuence on the future demand for
cultivated land. For example, substantial
increases in cereal yields and improved agricultural intensication could mean that the
increases in cultivated land required by Latin
America by 2050 may not be as large as predicted in Table 3. Second, although the existence of an EKC eect for agricultural
expansion in Latin America appears to be
highly sensitive to the model specication, the
relatively low estimated EKC turning point for
Latin America suggests that broader structural
economic changes and development in the
region could start to slow agricultural land
growth as well. The fact that political stability
may also be an important institutional inuence on slowing land expansion indicates that
political development may confer important
economic and environmental benets to the
region. Finally, the positive relationship
between the share of agricultural exports in
total merchandise exports and land expansion
in Latin America conrms that the increased
resource dependence of an economy may be
correlated with greater exploitation of its natural resource and land base. As will be discussed in the next section, a price-induced
resource boom may not always lead to
widespread economic benets, even in resourcedependent economies.
4. RESOURCE BOOMS AND ECONOMIC
GROWTH: IMPLICATIONS FOR LATIN
AMERICA
So far, we have established that agricultural
land expansion and natural resource exploitation by primary sector activities are fundamental to economic development in Latin
America. But, if per capita income is to be
sustained or increased in these economies,
especially with population increases, then any
depreciation of natural resources must be oset
by investment in other productive assets. This
implies that natural resource rents must be
reinvested in the economy by both managing
natural resources so as to maximize rents and
channeling those rents into productive investments elsewhere in the economy. Although it
would seem that the windfall prots generated
by resource price booms would be benecial to
this process, as we will discuss further below,
this may not be the case for resource-abundant

145

developing countries, especially in Latin

America.
In sum, even though an economy is endowed
with abundant natural resources, the country
may not necessarily be exploiting this natural
wealth eciently and generating productive
investments. Or, as Wright (1990, p. 666) has
succinctly put it: there is no iron law associating natural resource abundance with national
industrial strength. In fact, recent evidence
suggests that resource-abundant countries,
especially developing economies, may not be
beneting economically from this comparative
advantage. Many low-income and lower middle-income economies that can be classied as
highly resource dependent today also currently
display low or stagnant growth rates (Barbier,
1999). Crosscountry analysis has conrmed
that resource-abundant countriesi.e., countries with a high ratio of natural resource
exports to GDPhave tended to grow less
rapidly than countries that are relatively
resource poor (Sachs & Warner, 1997). Economies with a high ratio of natural resource
exports to GDP in 1971 also tended to have low
growth rates during the subsequent period
197189 (Sachs & Warner, 1995).
Such recent empirical evidence might be
considered surprising, given the commonly held
view that abundant natural resources ought to
be the basis for economic expansion for those
countries fortunate to have such a rich
endowment. For example, as historical precedent, empirical evidence clearly shows that the
origins of rapid industrial and economic
expansion in the United States over 18791940
were strongly linked to the exploitation of
abundant nonreproducible natural resources,
particularly energy and mineral resources
(Romer, 1996; Wright, 1990). In particular,
during 18801920, the intensity of US manufacturing exports in terms of nonreproducible
resources grew both absolutely and relative to
the resource-intensity of imports. There is also
evidence however that there were other factors
that made this historical situation in the United
States unique. For example, Wright (1990)
maintains that, over this era (a) the United
States was not only the worlds largest mineral
producing country but also one of the worlds
largest countries and markets; (b) high international transport costs and tari barriers for
manufactured goods compared to highly ecient and low cost domestic transportation
meant that the United States was a vast free
trade area for internal commerce and industrial

146

WORLD DEVELOPMENT

expansion that beneted from economic distance from the rest of the world; and (c)
because of the quantities of resources that were
available combined with the large internal
markets for goods, increasing investment in
basic technologies for extracting and processing
natural resources was highly protable. As
Wright (1990, pp. 665 & 661) suggests: the
abundance of mineral resources, in other
words, was itself an outgrowth of Americas
technological progress, and in turn, American producer and consumer goods were often
specically designed for a resource-abundant
environment.
It is doubtful however that the unique circumstances that allowed the United States to
achieve congruence between intensive
resource use and basic processing and manufacturing technologies over 18791940, and
thus attain rapid economic expansion, are
applicable to resource-abundant developing
economies today. For one, after 1940, this
unique congruence had clearly ended for the
United States, largely due to changes in the
international economy, even though the United
States still had abundant resources. As Wright
(1990, p. 665) points out:
the country has not become resource poor relative to
others, but the unication of world commodity markets (through transportation cost reductions and elimination of trade barriers) has largely cut the link
between domestic resources and domestic industries. . .
To a degree, natural resources have become commodities rather than part of the factor endowment of individual countries.

As some researchers have pointed out, the

changed international conditions during the
postwar era may have also aected the role of
primary-product export promotion as the
engine of growth for developing economies.
During this era, the main source of economic
growth in developing countries has not been
primary-product based exports but laborintensive manufactured exports (Findlay, 1996;
Findlay & Wellisz, 1993). 7
Not only are the conditions for congruence between resource abundance, technological progress and industrial expansion lacking
in developing economies today, but some have
also maintained that increased economic
dependence on resource exploitation may be
detrimental to innovation and growth. For
example, recent explanations of the limitations
of resource-based development have focused on

the poor potential for such development in

inducing the economy-wide innovation necessary to sustain growth in a small open economy. Matsuyama (1992) has shown that trade
liberalization in a land-intensive economy
could actually slow economic growth by
inducing the economy to shift resources away
from manufacturing (which produces learninginduced growth) toward agriculture (which
does not). Sachs and Warner (1995) also argue
that the relative structural importance of tradable manufacturing versus natural resource
sectors in an economy is critical to its growth
performance, i.e., when a mineral or oil-based
economy experiences a resource boom, the
manufacturing sector tends to shrink and the
nontraded goods sector tends to expand. This
phenomenon is often referred to in the literature as the Dutch disease eect. 8
Sachs and Warner (1999) have recently
examined evidence over 196094 for 11 major
Latin American economies to test the hypothesis that any natural resource booms occurring
in these countries my have had a positive impact
on their growth performance. 9 First, the
authors note that the main structural feature of
these economies is that they have remained by
and large exporter of primary commodities or
manufactured products based on these commodities. Second, they suggest that a signicant
resource boom occurred in only four of the 11
countries (Bolivia, Ecuador, Mexico and Venezuela), and mixed evidence of a boom in
another three (Chile, Colombia and Peru).
Sachs and Warner conclude however, that in
only one of these seven countries (Ecuador) did
a resource boom have a positive and lasting
eect on GDP per capita. In two countries
(Chile and Colombia) there appears to be no
eect of a resource boom on economic development, and in the remaining four cases
(Bolivia, Mexico, Peru and Venezuela), the
resource boom actually produced a negative
impact on GDP per capita. On balance,
resource booms appear to frustrate economic
growth in Latin America, most likely through a
Dutch disease eect.
If natural resource booms are not important
catalysts for economic development in poorer
countries, then perhaps the process of resource
exploitation and land expansion occurring in
these economies is not as economically benecial as it could be. That is, the structural economic dependence of a small open low or lower
middle income economy on exploiting its natural resource endowmentin particular its

AGRICULTURAL EXPANSION AND RESOURCE BOOMS

dependence on processes of agricultural land

expansionmay not be leading to sustained
and high rates of economic growth. This may
be occurring because natural resource assets,
including land, are not being managed so as to
maximize rents and/or whatever rents are being
generated in the economy are not being channeled into productive investments elsewhere in
the economy.
Brander and Taylor (1997, 1998) provide
some theoretical support for this perspective.
They note that overexploitation of many
renewable natural resourcesparticularly the
conversion of forests to agricultural land
occurs in developing countries if property
rights over a resource stock are hard to dene,
dicult to enforce, or costly to administer.
They demonstrate that opening up trade for a
resource-abundant economy with an open
access renewable resource may actually reduce
welfare in that economy. As the resourceabundant country has a comparative advantage
in producing the resource good, the increased
demand for the resource good resulting from
trade openness leads to greater exploitation,
which under conditions of open access leads to
declining welfare in the long run. Brander and
Taylor conclude that, as the problem lies with
the open access nature of exploitation in the
resource-abundant economy, then the rst-best
policy would be for the developing country to
switch to more ecient resource management
policy through simply establishing property
rights. As however, they acknowledge, and as
we discuss further in Section 5 below in the case
of Latin America, there are many policy and
institutional distortions that currently work
against such solutions in developing countries.
Consequently, Brander and Taylor (1997, p.
550) argue in favor of second best approaches such as the country imposing a modied
Hartwicks rule (see Hartwick, 1977) under
which an exporting country that experienced
temporary gains from selling a resource good
on world markets might re-invest those proceeds in an alternative asset.
As we discuss further in Section 5 with respect
to Latin America, current policies in resourceabundant developing economies appear not to
be ensuring that any resource rents earned are
re-invested eciently into other productive
assets in the economy. If this is the case, irrespective of what may happen to the countrys
terms of trade or commodity prices, any initial
economic boom associated with land conversion or increased resource exploitation is

147

invariably short-lived as the extra rents generated are eventually dissipated. Once the land
expansion and increased exploitation of new
resource reserves comes to an end, or poorer
quality land and resources are brought into
production, then some economic retrenchment
is inevitable. What we should therefore observe
is that economic development in a resourcedependent small open economy displays an
inherently boom and bust pattern.
Again, Brander and Taylor (1997) show that
a small, open and resource-abundant economy
that produces both a resource and a manufacturing good in the long run will have such a
pattern of development. That is, the economy
will experience early gains from trade, followed
by a period of declining utility. With the specic case of Latin America in mind, in which
raw materials are often inputs into semi-processed or processed exports, L
opez (1989) also
develops a two-good model of a resource-rich
open economy in which the open access
renewable resource serves as an input into an
enclave export processing sector. L
opez
shows that improvements in the terms of trade
increases the rate of open access resource
extraction and real income to increase in the
short-run, but inevitably permanent income
falls in the long run.
As argued throughout this paper, the classic
case of open access resource exploitation in
many developing countries is conversion of
forest to agriculture. If agricultural land
expansion in these small open economies is
associated with a boom and bust pattern of
economic development, then there are two
possible consequences. First, economies that
have increased their agricultural land base signicantly over the long run are likely to have
lower levels of GDP per capita then economies
that have tended to reduce their dependence on
agricultural land expansion. For the latter
countries, a shrinking agricultural land base
may be evidence that tradable manufacturing
and other dynamic sectors have become structurally more important in the economy relative
to natural resource sectors and that agriculture
itself has become a more capital-intensive,
productive and innovative sector. 10 Second,
for those countries that are dependent on
agricultural land expansion, further increases in
agricultural area will tend to produce only
modest increases in GDP per capita. Beyond a
certain point, additional increases in land
expansion will be associated with lower GDP
per capita, because of the boom and bust

148

WORLD DEVELOPMENT

pattern of resource-dependent development

described above.
A fairly straightforward way of empirically
verifying the above phenomenon is to empirically estimate a relationship between GDP per
capita and some measure of long-run agricultural expansion. For example, if the latter
indicator was some index, ait , then the above
hypotheses suggest that there may be a cubic
relationship between per capita income, Yit , and
this indicator of long run agricultural land
change:
Yit b0 b1 ait b2 a2it b3 a3it :

Note that b0 > 0, b1 < 0, b2 > 0, b3 < 0 and

jb1 j > b2 would imply that (i) countries with
increased long-run agricultural land area would
have lower levels of per capita income than
countries with decreased agricultural land area
and (ii) per capita income would tend to uctuate with long-run agricultural land expansion.
Eqn. (6) was estimated through employing a
panel analysis of tropical developing countries
over 196194. Per capita income, Yit , is again
represented by gross domestic product (GDP)
per capita in constant purchasing power parity

(1987 $). The indicator ait is an agricultural

land long-run change index, created by dividing
the current (i.e., in year t) agricultural land area
of a country by its land area in 1961.
The results of the analysis for all tropical
countries and Latin America are shown in
Table 5. For both regressions, the estimated
coecients are highly signicant and also have
the expected signs and relative magnitudes.
Thus the estimations provide some empirical
evidence that agricultural land expansion in
tropical countries and the Latin American
region conforms to a boom and bust pattern
of economic development. This is seen more
clearly when the regressions are used to project
respective relationships between long-run agricultural land expansion and GDP per capita,
which are displayed in Figure 1.
As indicated in the gure, an increase in
agricultural land expansion in the long run is
clearly associated with a lower level of per
capita income than decreasing agricultural land
area. For all tropical countries, the turning
point is a long run agricultural change index of
1.2. For Latin America the turning point is 1.3.
Although continued agricultural land expansion beyond these points does lead to a slight
increase in GDP per capita, this impact is

Table 5. Panel analysis of per capita income and long run agricultural expansion, 196194
Dependent variable: GDP per capita (PPP, constant 1987 $)a
Parameter estimatesb
All countries

Latin America

Explanatory variables

(N 1; 135)
14,393.37
(23.69)**

(N 401)
11,525.92
(3.62)**

Long run agricultural land area change index (ait )c

)24,293.31
()19.04)**
15,217.53
(11.182)**
)2,896.32
()6.59)**

)17,535.19
()2.55)**
12,156.88
(2.42)
)2,692.56
()2.20)**

168.01**
6,576.23**
6.85
0.368

151.13**
2,239.33**
1.68
0.145

One-way random eects

One-way random eects

a2it
a3it
F -test for pooled model
BreuschPagan (LM) test
Hausman test
Adjusted R2
Preferred model
a

Mean for all countries is $2,986, and for Latin America $3,675. PPP is purchase power parity.
t-Ratios are indicated in parentheses.
c
Mean for all countries is 1.18, and for Latin America 1.26.
*Signicant at 5% level.
**Signicant at 1% level.
b

AGRICULTURAL EXPANSION AND RESOURCE BOOMS

149

GDP per
capita
(constant
1987 $)
16000

14000

12000

10000

8000

6000

4000

2000

0
0

0.5

1.18 1.26

1.5

2.5

Agricultural land long-run change index

All Tropical

Latin America

Figure 1. Projections of agricultural land expansion and GDP per capita in tropical countries, 196194.

short-lived. For all tropical countries, per

capita income starts to fall once the land area
index reaches 2.3; for Latin America this occurs
sooner at an index of 1.7. Finally, the average
long-run agricultural land change index for all
tropical countries is 1.18, whereas it is 1.26 for
Latin America. As Figure 1 indicates, a comparison of these averages suggests that agricultural land expansion is associated with
slightly higher levels of per capita income than
in other tropical countries.
5. DO ECONOMIC POLICIES IN LATIN
AMERICA FURTHER AGRICULTURAL
LAND EXPANSION?
The above analysis suggests that agricultural
land expansion and resource exploitation in
Latin America may not contribute to economic
development because either insucient rents
are being generated from conversion and
depletion of natural resources or the rents
earned are not being invested in productive
assets or both. Ill-dened and lack of enforce-

ment of property rights for land create open

access conditions for exploiting land, which
may also pose a problem. There are three
possible ways in which current policies in the
region may exacerbate these problems. First,
existing land ownership and property right
regimes may encourage overuse and conversion
of land. Second, government policies may
contribute to unequal resource and land access,
which leads to increased frontier land conversion by poorer farmers. Third, recent structural
agricultural and trade liberalization reforms
may not generate the necessary incentives to
break the structural dependence of Latin
American economies on continued agricultural
land expansion. Specically, these largely
macroeconomic policy reforms are not aimed
at ensuring that any rents earned from resource
conversion and depletion are re-invested eciently into other productive assets in the
economy. Nor do these reforms address the
rst best policy solution identied by
Brander and Taylor (1997), namely encouraging more ecient resource management
through correcting distortions in property

150

WORLD DEVELOPMENT

rights and unequal access to resources, particularly land.

The general open access conditions of unoccupied forest land is now recognized as a key
condition underlying frontier agricultural
expansion in developing countries (Pearce,
Barbier, & Markandya, 1990). In Latin America, the problem is made worse by existing land
ownership rules. In particular, land titling regulations which essentially acknowledge forest
clearing as evidence of eective occupation and
ownership for both agriculture and livestock
raising have been documented as a major factor
in frontier agricultural conversion in Costa
Rica, Ecuador, Honduras, Panama and other
Latin American countries (Kaimowitz, 1995;
Mahar & Schneider, 1994; Peuker, 1992;
Southgate, Sierra, & Brown, 1991; Sunderlin &
Rodrgez, 1996). For example, in Costa Rica,
occupation of public lands has resulted in 60%
of farms lacking land title, and often competing
claims for land (Peuker, 1992). This has provided an incentive to undertake activities on the
land, such as clearing land of trees, which
clearly demonstrate possession. Title to land
can be obtained after 10 years of possession,
and a claimant can title up to 100 hectares (ha)
of land if the property is devoted to agriculture
and up to 300 ha if it is devoted to cattle
ranching. The process has proved to be highly
susceptible to fraud with respect to time of
occupation, the area of the land to be titled,
and the actual use of the land.
Land titling regulations can also aect the
initial opening up of unoccupied frontier
lands. It is often short-term extractive operationssuch as timber concessions, mining
concerns, large-scale commercial ranching and
farmingthat are more likely to be involved in
initial frontier development. Usually it is fairly
straightforward for governments to allocate
large tracts of frontier land to commercial
concerns and individual operators for extractive purposes, and often their activities receive
subsidies or other scal incentives of some
kind. Short-term land speculators may also be
encouraged in this way. Generally, the objective of these extractive and speculative operations is to maximize short-term resource rents;
long-term investment in frontier economic
development is not a major priorityparticularly if it is dicult to acquire long-term
property or use rights or to control illegal
occupation. As a consequence, once sucient
rents are extracted, land abandonment and
selling-o is common.

As noted above, however, once the frontier is

opened by large-scale activities, the lack of
secure property rights and general open access
conditions prevailing on the frontier inevitably
encourage rapid expansion of frontier agricultural activities by small-scale farming and
landless households in search of new lands. For
example, Schneider (1994) envisions this process as a perpetually moving frontier.
Extending roads into previously isolated forests
initiates the land grab, which is then accelerated
by the absence of property rights. As further
road-building and large infrastructure projects
are established in new areas, the process repeats
itself.
In Latin America, inequalities in wealth
between rural households also have an important impact on land degradation and deforestation processes. Such problems are exacerbated
by government policies that favor wealthier
households in markets for key resource, such as
land. First, poorer households are often unable
to compete with wealthier households in land
markets for existing agricultural land. The
result is two segmented land markets: the
wealthier rural households dominate the markets for better quality arable land, whereas the
poorer and landless households either trade in
less productive land or migrate to marginal
lands.
Second, although poorer households may be
the initial occupiers of converted forestland
they are rarely able to sustain their ownership.
As the frontier develops economically and
property rights are established, the increase in
economic opportunities and potential rents
makes ownership of the land more attractive to
wealthier households. Because of their better
access to capital markets, they can easily bid
current owners o the land, who in turn may
migrate to other frontier forest regions or
marginal lands.
For example, in Colombia distortions in the
land market prevent small farmers from
attaining access to existing fertile land (Heath
& Binswanger, 1996). That is, as the market
value of farmland is only partly based on its
agricultural production potential, the market
price of arable land in Colombia generally
exceeds the capitalized value of farm prots. As
a result, poorer smallholders and, of course,
landless workers cannot aord to purchase land
out of farm prots, nor do they have the nonfarm collateral to nance such purchases in the
credit market. In contrast, large landholdings
serve as a hedge against ination for wealthier

AGRICULTURAL EXPANSION AND RESOURCE BOOMS

households, and land is a preferred form of

collateral in credit markets. Hence the speculative and nonfarming benets of large land
holdings further bid up the price of land, thus
ensuring that only wealthier households can
aord to purchase land, even though much of
the land may be unproductively farmed or even
idled.
Similar to Colombia, tax and credit policies
in Brazil generally reinforce the dominance of
wealthier households in credit markets and the
speculative investment in land as tax shelters
(Mahar & Schneider, 1994). Because poorer
households on the frontier do not benet from
such policies, their ability to compete in formal
land markets is further diminished. This reinforces the sell out eect of transferring
frontier land ownership from poorer initial
settlers to wealthier and typically urban-based
arrivals, forcing the poorer households to drift
further into the frontier (Schneider, 1994).
According to L
opez (2003, p. 271) such policies in Latin America over the past 50 years are
symptomatic of the general economic policy
failure in the region that has
focused on the generation of an expensive and often
incoherent system of short-run incentives to promote
investment in physical capital. . . by undertaxing capital income and wasted in massive subsidies to the corporate sector in a futile eort to promote investment
and economic growth.

This has had two overall consequences on the

land degradation and deforestation process in
the region. First, as described above, the
resulting market and tax distortions promote
this process directly, in a deliberate strategy of
wasting natural resources as a way of enticing
investors (L
opez, 2003, p. 260). Second, Latin
American governments are dissipating scarce
revenues and nancial resources instead of
concentrating their eorts in raising enough
public revenues to nance the necessary
investment in human and natural capital and
the necessary institutional capacities to eectively enforce environmental regulations
(L
opez, 2003, p. 271).
Since 1990, many Latin American countries
have embarked on substantial reform of their
agricultural sector, including trade liberalization. These reforms have included tarication
with bound taris, eliminating quota restrictions, removing export taxes, and reducing or
eliminating the role of state trading agencies (de
Janvry, Key, & Sadoulet, 1997). An assessment

151

of eight Latin American and Caribbean

countries over 198594 highlights the major
impacts of these policy changes (Valdes,
1996). 11 The study found that the initiation of
trade liberalization reforms coincided unexpectedly with a drop in border prices of most
agricultural commodities below their long-term
trend, which combined with the appreciation of
exchange rates resulted in sharper than expected drops in real domestic producer prices.
Although the level of direct taxation of exports
has declined in some countries, taris on agricultural imports continue and are still higher
than those for nonagricultural imports. Trade
and price regimes still prevail that lead to
considerable overpricing of protected intermediate inputs and distortions in the product
output mix. The eects of the policy reforms on
net agricultural income transfers have been
somewhat uneven. 12 On the whole, the evidence suggests that government agricultural
policies across Latin America still tend to discriminate against exportable agricultural commodities.
Given the sector-wide eects of such major
economic reforms, it is dicult to assess the
resulting production responses of households,
except on a case study basis. A recent analysis
of Mexicos experience indicates some of the
complex linkages between agricultural pricing
policies, smallholder responses and expansion
of frontier activities.
Until the liberalization reforms in the early
1990s in Mexico, the livestock sector beneted
directly from preferential loans and subsidized
beef prices, and the agricultural sector from
maize price supports and subsidized fertilizer
inputs. A panel analysis for 197085 across all
Mexican states conrmed that planted agricultural area was highly correlated over this period
with the relative maize-fertilizer price ratio. In
addition, livestock numbers were positively
correlated with beef prices and credit disbursements, suggesting that the removal of
these subsidies would have a direct and negative eect on agricultural and livestock expansion (Barbier & Burgess, 1996). The potential
impact of agricultural policy reform on the
expansion of planted area and livestock numbers are good examples of the rst order, or
direct, eects of changes in pricing on the
incentives for frontier expansion and forest
conversion by rural households. The authors
suggest however, that there are also likely to be
some second order, or indirect, eects
resulting from economy-wide and sectoral

152

WORLD DEVELOPMENT

reforms that may produce opposite incentive

eects, potentially even outweighing the rst
order impacts. In particular, rural migration to
forested areas may increase as a result of the
impacts of economic reform on the returns and
value of existing agricultural land.
For example, as a result of the liberalization
reforms and structural changes as part of
joining the North American Free Trade
Agreement (NAFTA), Mexico is expected to
undergo a long transition to sustained economic growth accompanied by substantial
return migration to rural areas (Levy & van
Wijnbergen, 1992b). The positive inducement
to convert forest land, based on increases in the
rural labor force and falling rural wages, may
outweigh the incentives to reduce agricultural
and cattle expansion due to increasing real
GDP per capita (Barbier & Burgess, 1996).
The eect of the substantial reduction in the
producer price of maize in Mexico as a result of
the trade liberalization has been a reduction in
output and thus planted area. This has also
provoked a large fall in land values for rainfed
land, which is expected to decline to nearly onequarter that of irrigated land, thereby making
subsistence farmers, rainfed farmers who are
net sellers of maize, and landless rural workers
worse o (Levy & van Wijnbergen, 1992a).
Although subsistence farmers will benet from
lower consumer prices, they will be doubly
aected by the loss in value of their rainfed land
and in employment opportunities as day
laborers. The overall lack of employment and
income opportunities could induce rural
workers and subsistence farmers to migrate
toward frontier forest areas, or to convert
remaining forestland that is available to them
locally. These second order eects of trade
liberalization on deforestation could outweigh
the initial impacts of, say, a reduction in maize
producer prices on planted agricultural area
(Barbier & Burgess, 1996).
6. BREAKING THE CYCLE: A ROLE FOR
TARGETED ECONOMIC POLICIES?
The evidence presented in this paper has
indicated that the abundance of land and natural resources available in most Latin American countries does not necessarily mean that
exploitation of this natural wealth will lead to
sustained economic growth and benets for the
region. Although the focus of this paper was
mainly on the process of agricultural land

expansion, the results are probably generalized

across the natural resource sector of Latin
American economies.
First, the process of agricultural land
expansion appears to be driven mainly by the
structural characteristics of the agricultural
sectors of Latin America economiesrelative
low crop yields, high ratio of cropland to total
land area and a large share of agricultural
exports in overall merchandise exports.
Although there is some evidence that agricultural land expansion is likely to slow down as
Latin American economies develop and grow,
as long as the structural characteristics of
agriculture in the region remain fundamentally
unchanged, cultivated land area could increase
in Latin America by as much as 50% by 2050,
with 70% of the new land originating from
deforestation and wetland conversion (see
Table 3).
Second, the evidence from resource price
booms in Latin America suggests that there is
little indication that such booms generate sustained economic growth in the region. To the
contrary, they may lead to Dutch disease eects
whereby scarce resources are diverted from
more innovative industrial sectors to natural
resource exploitation. There is also evidence
that economies that have increased their agricultural land base signicantly over the long
run are likely to have lower levels of GDP per
capita then economies that have tended to
reduce their dependence on agricultural land
expansion. Continued land expansion may also
be correlated with a boom and bust pattern
of economic development.
Finally, current policies and reforms in Latin
America appear not to be overcoming the
structural characteristics in agriculture and
economic development that lead to the above
trends. That is, agricultural land expansion and
resource exploitation in Latin America may not
be contributing substantially to overall economic development because current policies are
failing to ensure that sucient rents are being
generated from conversion and depletion of
natural resources and that any rents earned are
re-invested eciently into other productive
assets in the economy. To overcome this
problem may require more targeted policies
aimed at correcting critical structural imbalances in Latin America economies that inhibit
the ecient exploitation of natural wealth for
sustained long-term development. 13
That is, as the recent policy reforms implemented in Latin American countries have very

AGRICULTURAL EXPANSION AND RESOURCE BOOMS

broad, economy-wide and sectoral impacts

with unknown (and possibly negative) aggregate impacts on land and resource use strategies
of rural households, it may be necessary to
complement these reforms with specic, targeted policies to generate direct incentives for
improved rural resource management. The
main aim of such policies would be to eliminate
distortions that reduce the economic returns of
existing smallholder agricultural lands, improve
the access of poorer rural households to credit
and land markets, and eliminate any remaining
policy biases that favor relatively wealthy
farmers and landowners.
For example, given the evidence presented in
this paper, policies that raise the returns on
existing agricultural land would more likely
lead to idle land being brought into production,
rather than more forestland conversion. In the
post-reform era, however, most agricultural
and livestock policies in Latin America are not
targeted, but instead have the overall objective of raising the production of key agricultural and livestock products, regardless of
whether this increased production comes from
existing or frontier agricultural land. If the
returns to both existing and convertible forest
land can be raised, then there will be little
change in the amount of forestland converted
to agriculture.
There is also substantial scope in many
Latin American countries to increase nonprice
transfers to the agricultural sector (Valdes,
1996; de Janvry et al., 1997). Such transfers
could reduce signicantly the incentives for
land degradation and forest conversion in
Latin America if they take the form of subsidies targeted to improve access by the rural
poor to credit, increased research and extension to disseminate information on the conservation and improvements of smallholder
land, and public investment in irrigation and
other infrastructure facilities to improve the
productivity and returns to existing smallholder land. For example, in Mexico there is
some evidence that an investment program in
land improvements to increase the productivity of rainfed land could potentially mitigate
the negative distributional implications of
NAFTA on the maize sub-sector (Barbier &
Burgess, 1996; Levy & van Wijnbergen,
1992a). Such a program could involve investments not only in irrigation infrastructure for
1.1 million ha of rainfed land but also in
drainage, land leveling, ditch-clearing and soil
conservation. Moreover, improvements in the

153

returns to existing agricultural areas and

smallholdings could lead to an expansion of
rural farm and o-farm employment opportunities, thus reducing migration by landless
and near-landless households to frontier areas.
This suggests that a land improvement
investment program for existing rainfed
farmers, particularly in states and regions
prone to high deforestation rates, could provide both direct and indirect incentives for
controlling deforestation by increasing the
comparative returns to farming existing
smallholdings as well as the demand for rural
labor.
Improving the extension of eective credit
markets and services to reach poor rural
households while continuing to eliminate subsidies and credit rationing that benet mainly
wealthier households may be important in
reducing resource degradation in many Latin
American countries. There is evidence that
economic reforms in Colombia, as well as in
other Latin American countries (e.g., Brazil,
Chile and Ecuador) may have signicantly
reduced credit subsidies to agriculture (Valdes,
1996). While this may have ended credit subsidies and rationing that had beneted mainly
wealthier farmers and select agricultural commodities, it did not increase rural credit for
land improvements, purchases, and investments
by poorer farmers. According to Heath and
Binswanger (1996), distortions in the credit
market have clearly formed a major part of the
incentives for rural households in Colombia to
migrate to both marginal upland areas and
equally fragile land in the forested AmazonOrinoco Basin.
As we have seen, institutional failures that
promote insecure tenure or ownership of land
also have an important inuence on resource
management decisions of rural households.
Tenure insecurity may mean that the incentives
to invest in land improvements are lacking, as
well as making it extremely dicult for farmers
to obtain formal credit for such improvements.
Thus an important inducement for many poor
smallholders to invest in improved land management is to establish proper land titling and
ownership claims on land currently occupied by
these smallholders. To improve land tenure
security in areas where frontier expansion is
occurring, it may be necessary to develop more
formal policies for smallholder settlement, such
as preferential allocation of public land, with
fully demarcated ownership and tenure rights
to smallholders.

154

WORLD DEVELOPMENT

NOTES
1. As a comparison, natural capital comprises only 5%
of wealth in North America, and 2% in the Pacic
OECD and Western Europe.
2. A major problem facing crosscountry analyses of
deforestation is the unreliability of any international
data set on forest cover, especially post-1990. The
United Nations Food and Agricultural Organization
(FAO), which has been the international agency responsible for compiling forest area data across all countries,
based its 1990 Global Forest Resource Assessment on
population growth projections in order to overcome an
inadequate forest database for some countries and
regions. This means that the FAO country forest cover
data are inappropriate for crosscountry analyses of
deforestation that use demographic factors as explanatory variables. This in turn means that crosscountry
analyses that employ the FAO forest cover data since
1990 and use demographic factors as explanatory
variables need to be treated with caution. An alternative
source of forest area data is the FAO Production
Yearbook. This includes data on forest, crop and
pasture land, but does not specify land area under
closed broadleaved forest. The data are drawn from
national government responses to surveys rather than
using primary data sources and are generally considered
to be less reliable than the Global Forest Resource
Assessment data. The inappropriateness of using the
FAO 1990 assessment based country forest cover estimates in crosscountry deforestation analyses has been
pointed out by Barbier and Burgess (1997, 2001),
Cropper and Griths (1994) and Deacon (1999). Given
the problems with recent FAO forest stock data highlighted above, some have argued that crosscountry
studies should concentrate on explaining agricultural
land expansion, Ait
Ait
1 rather than deforestation
across tropical countries (Barbier & Burgess, 2001).
3. Strictly speaking, deforestation is dened as (minus)
the percentage change in forested area, or
Fit
1
Fit =Fit
1 . Deforestation is however, clearly
related to the change in forest stock variable, Ft
Ft
1 ,
in Eqn. (1). In fact, various crosscountry analyses have
tended to use either specication as the dependent
variable to represent forest loss. To simplify notation,
Ft
Ft
1 is used in Eqn. (1) and subsequent equations as
a short-hand expression for deforestation.
4. Tropical countries are those countries with the
majority of their land mass lying between the tropics.
5. Including the three time-invariant institutional indices in a xed eects regression leads to collinear

regressors (Baltagi, 1995). As the institutional indices

are in themselves weighted country-specic dummy
variables, including the indices in an ordinary least
squares (OLS) regression will essentially imitate a xedeects model. Of course, the estimated coecients on the
institutional variables may also be including the inuence of other slow-changing factors that vary across
countries.
6. The failure of the property rights index to be
signicant may reect the fact that this variable indicates
the degree of protection of private property rights across
all sectors of the economy rather than the security of
land tenure in the agricultural sector. The failure of the
corruption index to be signicant may be the result of
two countervailing inuences. First, increased corruption is associated with greater rent-seeking behavior,
which in turn may induce expansion of large-scale
commercial agricultural investments, leading to greater
land conversion. On the other hand, L
opez (1998) has
argued that reduced corruption and improved bureaucratic eciency may actually facilitate the implementation of land and credit policies that stimulate a race for
property rights to convert forest and other common
resource land to agriculture.
7. From their case study analysis of ve open developing economies, Findlay and Wellisz (1993) conclude
that over the postwar era it was economies with
relatively no resources, such as Hong Kong, Singapore
and Malta, that were among the earliest and most
successful exporters of labor-intensive manufactures. In
contrast, resource-rich Jamaica and the Philippines have
done relatively poorly, whereas Indonesia and Malaysia
have done comparatively better by balancing primary
exports with rapid expansion of labor-intensive manufactures.
8. Originally, the Dutch disease phenomenon was
associated with the macroeconomic implications of an
economys over-dependence on a single, traded natural
resource sector (e.g., oil), which emphasized the enclave
character of the sector as the predominant source of
foreign exchange availability (e.g., see Neary, van
Wijnbergen, & van Wijnbergen, 1986). As the consequence of a resource price boom (e.g. oil price shock),
expansion of the resource-based sector would be accompanied by a change in the real exchange rate, and the rest
of the economy would decline relatively. The more
recent treatments of the Dutch disease phenomenon,
such as by Matsuyama (1992) and Sachs and Warner
(1995) discussed here, focus less on the economic
implications of a resource boom via real exchange rate

AGRICULTURAL EXPANSION AND RESOURCE BOOMS

movements but via internal economic distortions caused
by the shift of resources from a more innovative sector
(e.g., manufacturing) to a less innovative sector (e.g.,
agriculture, minerals). This latter representation of the
Dutch disease is more appropriate for characterizing a
small open economy, in which real exchange rate
determination is not considered.
9. The countries are Argentina, Bolivia, Brazil, Chile,
Colombia, Ecuador, Mexico, Paraguay, Peru, Uruguay
and Venezuela.
10. In the small open economy model of Brander and
Taylor (1997), if the country specializes in the manufacturing of goods in the long run, it gains unambiguously
from trade.

11. The countries are Argentina, Brazil, Chile, Colombia, the Dominican Republic, Ecuador, Paraguay and
Uruguay.

155

12. These transfers were measured as the change in

value-added resulting from price and trade interventions
on inputs and nal products, measured at the actual
level of production, plus the net value of nonprice
related transfers (Valdes, 1996). For those countries with
substantial income outows, input subsidies and nonprice transfers (investment in irrigation, research and
extension, credit subsidies and other government expenditures) did not compensate, or compensated very little
for the substantial income outows.
13. Such targeted policies may have also the scope to
reduce rural poverty substantially. See de Janvry et al.
(1997) for further discussion. The proposed investment
strategy would also represent a substantial shift in the
overall policy framework in Latin America away from
futile eorts to promote physical and nancial capital
accumulation, which have beneted mainly of wealthy
urban investors, at the expense of protecting and
enhancing overall natural and human capital, which
benet mainly the rural poor. See L
opez (2003) for
further discussion.

REFERENCES
Alston, L. J., Libecap, G. D., & Mueller, B. (1999).
Conicts and land use: the development of property rights and land reform in the Brazilian Amazon frontier. Ann Arbor: University of Michigan
Press.
Alston, L. J., Libecap, G. D., & Mueller, B. (2000). Land
reform policies, the sources of violent conict, and
implications for deforestation in the Brazilian Amazon. Journal of Environmental Economics and Management, 39(2), 162188.
Anderson, L. E., Granger, C. W. J., Eustaquio, J. R.,
Weihold, D., & Wunder, S. (2002). The dynamics of
deforestation and economic growth in the Brazilian
Amazon. London: Edward Elgar.
Andreoni, J. R., & Levinson, A. (2001). The simple
analytics of the environmental Kuznets curve. Journal of Public Economics, 80, 269286.
Antle, J. M., & Heidebrink, G. (1995). Environment and
development: theory and international evidence. Economic Development and Cultural Change, 43(3), 603
625.
Baltagi, B. H. (1995). Econometric analysis of panel data.
Chichester: John Wiley.
Barbier, E. B. (1999). Poverty, environment and development. In J. C. J. M. van den Bergh (Ed.), The
handbook of environmental and resource economic
(pp. 731744). London: Edward Elgar.
Barbier, E. B. (2002). Institutional constraints and
deforestation: an application to Mexico. Economic
Inquiry, 40(3), 508519.
Barbier, E. B., & Burgess, J. C. (1996). Economic
analysis of deforestation in Mexico. Environment and
Development Economics, 1(2), 203240.

Barbier, E. B., & Burgess, J. C. (1997). The economics of

tropical forest land use options. Land Economics,
73(2), 174195.
Barbier, E. B., & Burgess, J. C. (2001). The economics of
tropical deforestation. Journal of Economic Surveys,
15(3), 413421.
Beck, T., Levine, R., & Loayza, N. (2000). Finance and
the sources of growth. Journal of Financial Economics, 58(October), 261300.
Brander, J. A., & Taylor, M. S. (1997). International
trade and open-access renewable resources: the small
open economy. Canadian Journal of Economics,
30(3), 526552.
Brander, J. A., & Taylor, M. S. (1998). Open access
renewable resources: trade and trade policy in a twocountry model. Journal of International Economics,
44, 181209.
Chomitz, K. M., & Gray, D. P. (1996). Roads, land
markets and deforestation: a spatial model of land
use in Belize. The World Bank Economic Review,
10(3), 487512.
Cropper, M., & Griths, C. (1994). The interaction of
population growth and environmental quality.
American Economic Review. AEA Papers and Proceedings, 84(2), 250254.
Deacon, R. T. (1994). Deforestation and the rule of law
in a cross-section of countries. Land Economics,
70(4), 414430.
Deacon, & Robert, T. (1999). Deforestation and ownership: evidence from historical accounts and contemporary data. Land Economics, 75(3), 341359.
de Janvry, A., Key, N., & Sadoulet, E. (1997). Agricultural and rural development policy in Latin America:

156

WORLD DEVELOPMENT

new directions and new challenges. Working

Paper No. 815. Department of Agricultural and
Resource Economics, University of California at
Berkeley.
Ehui, S. K., & Hertel, T. W. (1989). Deforestation and
agricultural productivity in the C^
ote dIvoire. American Journal of Agricultural Economics, 71(August),
703711.
Findlay, R. (1996). Modeling global interdependence:
centers, peripheries and frontiers. American Economic Review, 86(2), 4751.
Findlay, R.& Wellisz, S. (Eds.). (1993). The political
economy of poverty, equity and growth: ve small open
economies, a World Bank comparative study. New
York: Oxford University Press.
Fischer, G., & Heilig, G. K. (1997). Population momentum and the demand on land and water resources.
Philosophical Transactions of the Royal Society Series
B, 352(1356), 869889.
Food and Agricultural Organization (FAO) (1993).
Forest resources assessment 1990: tropical countries.
Rome: FAO.
Food and Agricultural Organization of the United
Nations (FAO) (1995). World agriculture:towards
2010an FAO study. Rome and New York: FAO
and John Wiley & Sons.
Food and Agricultural Organization of the United
Nations (FAO) (1997). State of the Worlds Forests
1997. Rome: FAO.
Food and Agricultural Organization (FAO) (2001).
Forest resources assessment 2000: preliminary results.
Rome: FAO.
Godoy, R., Jacobson, M., De Castro, J., Aliaga, V.,
Romero, J., & Davis, A. (1998). The role of tenure
security and private preference in neotropical deforestation. Land Economics, 74(2), 162170.
Hartwick, J. (1977). Intergenerational equity and the
investing of rents from exhaustible resources. American Economic Review, 6, 972974.
Heath, J., & Binswanger, H. (1996). Natural resource
degradation eects of poverty and population
growth are largely policy-induced: the case of
Colombia. Environment and Development Economics,
1, 6583.
Kaimowitz, D. (1995). Livestock and deforestation in
Central America in the 1980s and 1990s: a policy
perspective. EPTD Discussion Paper, No. 9, Environment and Production Technology Division, International
Food
Policy
Research
Institute,
Washington, DC.
Koop, G., & Toole, L. (1999). Is there an environmental
Kuznets curve for deforestation? Journal of Development Economics, 58(July), 231244.
Levine, R., Loayza, N., & Beck, T. (2000). Financial
intermediation and growth: causality and causes.
Journal of Monetary Economics, 46(1), 3177.
Levy, S., & van Wijnbergen, S. (1992a). Maize and the
free trade agreement between Mexico and the United
States. The World Bank Economic Review, 6(3), 481
502.
Levy, S., & van Wijnbergen, S. (1992b). Transition
problems in economic reform: agriculture in the
Mexico-US free trade agreement. Policy Research

Working Paper No. 967. Washington, DC: The

World Bank.
L
opez, R. (1989). Exchange rate determination in
natural resource-rich economies. Revista de Analisis
Economico, 4(1), 91105.
L
opez, R. (1998). Where development can or cannot go:
the role of poverty-environment linkages. In B.
Pleskovic & J. E. Stiglitz (Eds.), Annual Bank
conference on development economics 1997 (pp. 285
306). Washington, DC: The World Bank.
L
opez, R. (2003). The policy roots of socioeconomic
stagnation and environmental implosion: Latin
America 19502000. World Development, 31(2),
259280.
Mahar, D., & Schneider, R. R. (1994). Incentives for
tropical deforestation:some examples from Latin
America. In K. Brown & D. W. Pearce (Eds.), The
causes of tropical deforestation. London: University
College London Press.
Matsuyama, K. (1992). Agricultural productivity, comparative advantage, and economic growth. Journal of
Economic Theory, 58, 317334.
McConnell, K. E. (1997). Income and the demand for
environmental quality. Environment and Development
Economics, 2, 383399.
Neary, J., van Wijnbergen, P., & van Wijnbergen, S.
(Eds.). (1986). Natural resources and the macro
economy. Cambridge, MA: MIT Press.
Nelson, G. C., Harris, V., & Stone, S. W. (2001).
Deforestation, land use, and property rights: empirical evidence from Darien, Panama. Land Economics,
77(2), 187205.
Panayotou, T. (1995). Environmental degradation at
dierent stages of economic development. In I.
Ahmed & J. A. Doeleman (Eds.), Beyond Rio: the
environmental crisis and sustainable livelihoods in the
Third World. London: Macmillan Press.
Pearce, D. W., Barbier, E. B., & Markandya, A. (1990).
Sustainable development: economics and environment
in the Third World. London: Edward Elgar.
Peuker, A. (1992) Public policies and deforestation: a case
study of Costa Rica. Latin America and the Caribbean Technical Department, Regional Studies Program, Report No. 14. Washington, DC, The World
Bank.
Romer, P. M. (1996). Why, Indee, in America? Theory,
history, and the origins of modern economic growth.
American Economic Review, 86(2), 202212.
Sachs, J. D., & Warner, A. M. (1995). Natural resource
abundance and economic growth. National Bureau of
Economic Research Working Paper No. 5398,
December.
Sachs, J. D., & Warner, A. M. (1997). Fundamental
sources of long-run growth. American Economic
Review, 87(2), 184188.
Sachs, J. D., & Warner, A. M. (1999). The big push
natural resource booms and growth. Journal of
Development Economics, 59, 4376.
Schneider, R. R. (1994). Government and the economy on
the Amazon frontier. Latin America and the Caribbean Technical Department, Regional Studies Program, Report No. 34. Washington, DC: The World
Bank.

AGRICULTURAL EXPANSION AND RESOURCE BOOMS

Shak, N. (1994). Economic development and environmental quality: an econometric analysis. Oxford
Economic Papers, 46(October), 757773.
Southgate, D. (1994). Tropical deforestation and agricultural development in Latin America. In K. Brown
& D. W. Pearce (Eds.), The causes of tropical
deforestation: the economic and statistical analysis of
factors giving rise to the loss of the tropical forests
(pp. 134145). London: University College London
Press.
Southgate, D., Sierra, R., & Brown, L. (1991). The
causes of tropical deforestation in Ecuador: a statistical analysis. World Development, 19(9), 11451151.
Stokey, N. L. (1998). Are there limits to growth?
International Economic Review, 39, 131.

157

Sunderlin, W. D., & Rodrgez, J. A. (1996). Cattle,

broadleaf forests and the agricultural modernization
law of Honduras. CIFOR Occasional Paper No. 7.
Jakarta: Centre for International Forestry Research.
Valdes, A. (1996). Surveillance of agricultural price and
trade policy in selected Latin American countries at
the time of major policy reform. Technical Department, Latin American and Caribbean Region.
Washington, DC: The World Bank.
World Bank (1997). Expanding the measure of wealth:
indicators of environmentally sustainable development.
Washington, DC: The World Bank.
Wright, G. (1990). The origins of American industrial
success, 18791940. American Economic Review,
80(4), 651668.