The agriculture-urbanization nexus in economics: Three empirical studies

Xiaofei Li, Purdue University

Abstract

This dissertation empirically examines the spatial and temporal relationship between agriculture and urbanization. It comprises of three empirical studies investigating the relationship from three different aspects. The first empirical study quantitatively assesses how agricultural land quality has determined the spatial distribution of the urban population across the globe, and how that relationship has changed over time. It is an empirical validation of a conventional view in the narrative economic history literature, which maintains that the geographical distribution of urbanization was highly dependent on local agricultural land quality in early periods, whereas this relationship is much weaker in modern times. The study is based on global 0.5-degree grid cell data, and use the spatial negative binomial regression model to relate the urban population to agricultural suitability. It controls for other physical geographic factors, spatial interdependence between urban areas, and regional fixed effects. The model is estimated at various different points in time during the period ranging from 1000 CE to 2000 CE. The estimated coefficient of agricultural suitability first increased in early history (before 1500 CE), and then decreased over time. The inverted U-shaped pattern contradicts the predominant view in the literature, and is invariant under various robustness checks. It provides strong empirical evidence supporting the declining importance of fertile agricultural lands in determining locations of urbanization over time. The second study examines how the role of agriculture in urbanization changes with economic openness. The literature offers conflicting views about the effect of agricultural productivity on a countries' urbanization level. Theoretical models suggest that the effect alters with economic openness; that the relationship between agriculture and urbanization is positive for closed economies and negative for open economies. However, until now the empirical validation for the interaction effect with economic openness is almost non-existent. Based on a panel of 97 developing countries spanning the period 1992-2012, this study specifies a panel smooth transition regression model in which the coefficient for agricultural productivity is allowed to vary as a logistic smooth function of a countries' degree of economic openness (measured by the international trade to GDP ratio). The results show that for economically closed countries the effect of agricultural productivity on urbanization is statistically significantly positive. As the degree of a country's trade openness increases, the effect declines. For the most open countries in the world, the effect is not statistically significantly different from zero. The third case study empirically assesses whether the rapid urbanization in China has affected agricultural production based on panel data for 31 Chinese provinces from 2000-2011. There has been considerable debate on whether China's rapid urbanization threatened its agricultural production and subsequent food security, but so far most studies have used computable general equilibrium techniques for simulation and forecasting purposes, rather than providing a strong empirical test using historical, observed data. This study uses spatial panel models to examine whether the Chinese provinces with higher urbanization rates have experienced slower growth in agricultural output. The empirical results show no statistically significant evidence that agricultural growth was negatively associated with urbanization across provinces. Instead, we find agricultural growth was negatively correlated with initial yield levels in the provinces, positively correlated with land intensity, and negatively affected by natural disasters.

Degree

Ph.D.

Advisors

Waldorf, Purdue University.

Subject Area

Economics|Agricultural economics

COinS