4. Suitability for Rainfed and Irrigated Cereals Soil Fertility Constraints Source: Global Agro-Ecological Zones (GAEZ), 2000 – Food and Agriculture Organization of the United Nations (FAO) with the International Institute for Applied Systems Analysis (IIASA)
8. (trendline and confidence bands at the 99% level) Institutions data from Kaufmann, Kraay and Mastruzzi 2003, Governance Matters III: Governance Indicators for 1996-2002, World Bank.
I approach this these topics using recent… Briefly considering these debates helps illustrate the areas I hope to teach By 1990s, there was increasing consensus among many economists that development theory and policies simply were not working Harsh critiques of theory and policy, made all the more powerful because many of them came from ‘insiders’ For example, around the year 2000, three ‘shots across the bow’ of development theory and policy… (SLIDE 2)
[talk from slide] IMF economist Joseph Stiglitz “what I learned at the world Economic Crisis" in the New Republic , World Bank water scarcity expert Maggie Black ‘ The No-Nonsense Guide to Economic Development ” And World Bank economist William Easterly’s “ The Elusive Quest for Growth: Economist’s Adventures and Misadventures in the Tropics mounting evidence of failure led to renewed openness among economists to numerous factors formerly black-boxed as ‘exogenous’ especially geographic factors long ignored by economists, and social factors, often because thought too hard to quantify. Geographic factors including highly detailed measures of geographic factors such as agricultural potential (soils, climate, water scarcity), or transport cost limitations This has also been partly driven by creation of environmental datasets for environmental studies that have then used by economists. Perhaps the most influential of this research has been the work of Jeffrey Sachs and former colleagues at the Center for International Development (CID) at Harvard Seems like (and criticized by some geographers as) ‘Old Fashioned Geography’, although with greatly enhanced coverage and detail of datasets, and much more nuanced view of knock-on effects (amplifications) (e.g., disease distributions affect life expectancy which in turn affects investment in education and in turn development) Mediated by social factors
SLIDE 3 Here we have Global Frost data Economic study of positive effects of frost on agriculture (pest control, soil development)- linking this to development data. Similarly, have huge datasets based on combined global, yet extremely small scale [thousands of observations at 1 degree by 1 degree scales or smaller] data on numerous factors such as precipitation, evapotranspiration, frost days, terrains slopes, soil depth, soil fertility, soil chemistry, soil drainage, and soil texture
…top image (bottom image) together shows suitability for different crops. What is especially novel is the use by economists of these types of geographic data in their regressions on economic growth
More – pop density near and far from ocean navigable rivers and coast, Global GDP density per square kilometer (Yale G-Econ project) Measures of Institutions development of the first truly global data on social and institutional factors such as education levels, measures of institutions and legal systems, and social values Example: Massive World Bank dataset ‘Governance Matters’ 199 countries and territories several hundred individual variables drawn from 25 separate data sources constructed by 18 different organizations. Factor analysis into Six dimensions of institutions voice and accountability political stability government effectiveness regulatory quality rule of law control of corruption
However, as these datasets have grown ever more detailed, covering more factors, and often with global coverage, they have often raised more questions than they have answered. One reason is because these factors show what is, for the social sciences, unusually robust levels of covariation. For example, consider a comparison of the Governance Matters data mentioned in the previous slides with a simple measure of Gross National Income. We plot the Institutional data on the vertical axis and GNI on the horizontal axis. The trendline with confidence bands at the 99 percent level is shown.
There are numerous similarly close correlations between many geographic factors and development indicators, as well as between measures of institutions and geography. Wide enough discussion to become known as the ‘Institutions vs. Geography’ debate among economists Two papers especially influential, especially because of ingenious attempts to sort out covariation between geography and institutions using a natural experiment and reverse causality between institutions and wealth using instrumental variables
During early period when I was researching geographic, political and economic development theories, a renewed debate on the relationship of these began among development economists. This debate concerned the relevant importance of social factors such as social capital or quality of government against geographic factors such as distributions of disease, agricultural productivity, or transport costs. This renewed ‘geography versus institutions’ debate raises both abstract and empirical questions central to geography, :
[READ SLIDE] -------------------------------- Too much to cover now, but an example gives an idea of the general way I approach these topics Specifically, let’s consider how old fashioned world regional geography is still crucial for better interpreting ‘cutting edge’ comparative and econometric development studies Modern econometric and comparative studies often rely on large datasets for statistical analysis of how various factors seem to influence each other Let’s consider two of these datasets.
The first is designed to be a better measure than simple measures of Gross Domestic Product (GDP) Vertical = ratio of industrialized manufactures in a country (textiles, metals manufactures, chemicals, paper and paper products) against natural resource intensive agricultural and mining products (basic metals, foods, beverages tobacco, wood and wood products). Horizontal = Percent –non-agricultural population divides them into six categories: IC ( I ndustrialized, C ore industrial manufactures exporting) IN ( I ndustrialized, N atural resource intensive exporting) PIC ( P artially I ndustrialized, C ore industrial manufactures exporting) PIN ( P artially I ndustrialized, N atural resource intensive exporting) LIC ( L east I ndustrialized, C ore industrial manufactures exporting) LIN ( L east I ndustrialized, N atural resource intensive exporting)
A second example is a vast multi-year, global survey of subtle differences in social views possibly related to development outcomes covers sixty five societies representing 75% of the world’s population covers a wide range of attitudes and beliefs on political, social and religious questions Factor analysis of these scores by Inglehart (2000) revealed two axes of strong variation across countries. One axis, ‘traditional/secular rational’, relates to attitudes towards such areas as religion, family size, family ties, deference to authority, acceptance of military rule, and national pride. A second axis, ‘survival/self expression’, taps into the intergenerational shift from a traditional focus on security against ‘increasing emphasis on self expression, subjective well-being, and quality of life’ and attitudes towards such things as the environment and gender.
These are the types of large datasets being used. When regression analysis is run, the cases disappear. Also, danger of Type 1 errors – where we are overconfident in our arguments because there seem to be many cases that confirm them. Lets think about that second problem, and then see how it might relate to One type of spatial autocorrelation is where the factor being measured does not match up to the borders it is measured by. For example, imagine we are comparing units of Blue against Units of Gray. , consider this square (SLIDE 24)
Approximately 195 countries
World regional geography has long sought to divide world into regions based on historical cultural interaction, agricultural ecumenes Religion and language Oversimplified – What a geographer and historian together called the Myth of Continents: A Critique of Metageography The most thorough recent consideration of a global division based on historical cultural influence and known trade and economic ties is that by Lewis and Wigen (1997). After several hundred pages of analysis, Lewis and Wigen decide on the contemporary map of world regions shown above. Although in appearance this map is similar to world regional divisions in older atlases, the borders above are based on very careful consideration of inter-regional political, cultural, and linguistic ties, migration and economic interaction and influence.
the same data grouped by world region (with the exception of group D): The seven clusters outlined are A: Japan, Korea, Hong Kong and Singapore; B: most of Western Europe, the United States and Canada; C: Eastern Europe and the Mediterranean, D: wealthy to semi-wealthy natural resource intensive countries, such as Australia, New Zealand, Argentina, Iceland, Greenland, Norway, South Africa, and Russia; E: Latin American Countries, F: Central and South Asian economies with manufacturing exports i.e. China, Bangladesh, Nepal, Pakistan, Thailand, India; and G: Sub-Saharan African countries. and the degree to which it is likely that most econometric development studies suffer from Type-I errors in analysis.
In summary: ‘Old fashioned’ world regional geography shows up in ‘cutting edge’ econometric data Seeing the data in this way allows for a clearer picture of what is being compared against what in econometric studies. This is usually obscured by common statistical techniques. This is useful for at least two reasons: (Click)
Overall, my research strives to show how relatively simple visual methods of data analysis can help in the interpretation of (and highlight problems with) complex debates in development studies. I hope to use these methods to explore spatial development studies as well as teach the methods themselves.