International comparison and evaluation of agricultural productivity growth

Graduation date: 2009Long-run food price changes respond to long-run changes in food demand, quasi-fixed and variable production factors, and agricultural productivity. Recent global food-price volatility puts a renewed spotlight on the trends and sources of any agricultural productivity growth. Because food prices’ most critical effects are on human hunger, it is especially important to examine productivity conditions in newly emerging economies. Two such economies evaluated in this dissertation are Indonesia and Brazil.\ud \ud As agriculture’s share of the Indonesian economy declines in the post-Green Revolution era, government attention increasingly has shifted to the industrial sector and away from agriculture. In light of these trends, we use province-level data and a multi-output frontier distance function approach to estimate product-specific productivity change on Indonesian farms, decomposing it into its technical-change and efficiency-change components. We find, at a 1.4% annual growth rate, that technical change has been modest and that little of this growth can be ascribed to government research efforts. Furthermore, average farm efficiency relative to the best-practice frontier has declined 0.4% per annum, so that mean productivity has risen by only 1% per year. Over-centralization of the agricultural research system may partly account for this poor performance. \ud \ud Brazil now is the largest coffee, sugar, and fruit juice producer, second-largest soybean and beef producer, and third-largest corn and broiler producer. It has overtaken the U.S. in poultry exports, nearly matches the U.S. in soybean exports, and dominates global trade in frozen orange juice. To test and better understand these advances, we draw on decennial farm censuses to examine technical change and efficiency in Brazilian agriculture. Our approach is to estimate a stochastic, multi-product, output distance frontier, using a translog functional form and data disaggregated to the micro-region (sub-state) level. Using two consecutive decennial farm censuses, we combine state-level Fisher productivity-change indexes with state-level translog distance function estimates of growth technical efficiency to impute state-level technical shifts. We find, leading up to the soon-to-be-released 2006 agricultural census, that Brazil’s multi-factor productivity growth rate between 1985 and 1996 was 20.2%. Mean state-level growth efficiency was 91.2%, implying the production frontier expanded 22.2% over the reference time period