Data and Methods
Wiebe
et al. (2000) examined the impact of resource quality
on the productivity of agricultural land, using for the
first time recent global data on soils, climate, and land
cover. They began with data
developed by Hari Eswaran and others at USDA's Natural
Resources Conservation Service, who combined the Food
and Agriculture Organization's Digital
Soil Map of the World and associated soil characteristics
(e.g. slope, depth, and salinity) with spatially referenced
long-run average temperature and precipitation data to
establish nine land quality classes in terms of their
suitability for agricultural production.
Wiebe
et al. (2000) then overlaid these land quality classes
with political boundaries and global
land-cover data generated by the U.S. Geological Survey
from satellite imagery with a resolution of one kilometer.
They focused on cropland areas identified according to
the International Geosphere-Biosphere Programme land cover
classification scheme. The result is a land quality indicator
based on the share of each country's cropland that is
found in the three best quality classes. Countries where
this share exceeds the median value for their region are
identified as having good soils and climate; those with
less than the median are identified as having poor soils
and climate.
This
static measure, based on cross-country differences in
inherent soil and climate characteristics, supplements
existing time-variant quality indicators such as the percentage
of agricultural land that is cropped (or irrigated) and
long-term average or annual rainfall. To better capture
this last effect, we also developed a high-resolution
measure of annual rainfall by aggregating and overlaying
monthly precipitation
data on a 0.5-degree grid from the University of East
Anglia's Climatic Research Unit with national boundaries
and cropland as described above. The result is a country-specific,
time-variant measure of annual rainfall on cropland.
The dependent variable in our analysis is the productivity of agricultural
labor, measured as the value of total agricultural production (the
sum of price-weighted quantities of all agricultural commodities,
expressed in international dollars, after deductions for feed and
seed) per agricultural worker. Other variables are drawn from the
Food and Agriculture Organization
and the World Bank,
and include country-level indicators of agricultural land, tractors,
livestock, and fertilizer, as well as measures of the quality of
labor, the institutional environment, and infrastructure. The data
are combined in an econometric analysis of 110 countries over the
period 1961-97. Additional detail is provided in Wiebe
et al. (2000).
To estimate yield impacts of soil erosion, we used data from 90
original field studies that report quantitative yield results from
experiments conducted in the United States and Canada between 1939
and 1999. Mean erosion-induced yield declines were estimated for
each crop and soil order.
The total production decline resulting from erosion on the soils
considered here was estimated by multiplying area and annual yield
loss per hectare and summing across soil orders for each crop in
the United States and Canada.
To further aggregate impacts across crops, we multiplied these
annual production loss estimates by USDA baseline price projections
for 2000 (USDA) to estimate the total gross economic value of erosion-induced
loss of production. Additional detail is available in den Biggelaar
et al. (2001).
|
