Accounting for Cross-Country Income Differences Francesco Caselli outline
Abstract
Why are some countries so much richer than others? Development Accounting is a first-pass
attempt at organizing the answer around two proximate determinants: factors of production
and efficiency. It answers the question “how much of the cross-country income variance can
be attributed to differences in (physical and human) capital, and how much to differences
in the efficiency with which capital is used?” Hence, it does for the cross-section what
growth accounting does in the time series. The current consensus is that efficiency is at least
as important as capital in explaining income differences. I survey the data and the basic
methods that lead to this consensus, and explore several extensions. I argue that some of
these extensions may lead to a reconsideration of the evidence.
Contents 1 Introduction 2 The Measure of Our Ignorance 2.1 Basic Data 2.2 Basic Measures of Success 2.3 Alternative Measures Used in the Literature 2.4 Sub-samples 3 Robustness: Basic Stuff 3.1 Depreciation Rate 3.2 Initial Capital Stock 3.3 Education-Wage Profile 3.4 Years of Education 1 3.5 Years of Education 2 3.6 Hours Worked 3.7 Capital Share 4 Quality of Human Capital 4.1 Quality of Schooling: Inputs 4.1.1 Teachers’ Human Capital 4.1.2 Pupil-Teacher Ratios 4.1.3 Spending 4.2 Quality of Schooling: Test Scores 4.3 Experience 4.4 Health 4.5 Social vs. Private Returns to Schooling and Health 5 Quality of Physical Capital 5.1 Composition 5.2 Vintage Effects 5.3 Further Problems with K 6 Sectorial Differences in TFP 6.1 Industry Studies 6.2 The Role of Agriculture 6.3 Sectorial Composition and Development Accounting 7 Non-Neutral Differences 7.1 Basic Concepts and Qualitative Results 7.2 Development Accounting with Non-Neutral Differences 8 Conclusions
8 Conclusions Development accounting is a powerful tool to getting started thinking about the sources of income differences across countries. As of now, the answer to the development-accounting question — do observed differences in the factors employed in production explain most of the cross-country variation in income — is: no, way no. This negative answer is robust to attempts to improve the measurement of human capital by allowing for differences in the quality of schooling and in health status of the population; to attempts to account for the age composition of the capital stock; to sectorial disaggregations of output; and to several other robustness checks. On the other hand, incomplete knowledge about certain key parameters that describe the relationship between inputs and outputs implies that the jury should be treated as being still out. For one thing, depending on the elasticity of substitution between capital of different 61 types, the observed wild heterogeneity in the composition of the capital stock by type of equipment could turn out to be a key proximate determinant of income differences. For another, depending on the elasticity of substitution between physical and human capital, we may find that all is needed for these factors to explain a large fraction of income inequality is a departure from Cobb-Douglas. Disaggregating the government sector out of the data may also potentially reduce the unexplained component of GDP. There is no deep reason why we should not be able to make progress on these three fronts, so that my assessment of the future of this research enterprise is optimistic.
Contents 1 Introduction 2 The Measure of Our Ignorance 2.1 Basic Data 2.2 Basic Measures of Success 2.3 Alternative Measures Used in the Literature 2.4 Sub-samples 3 Robustness: Basic Stuff 3.1 Depreciation Rate 3.2 Initial Capital Stock 3.3 Education-Wage Profile 3.4 Years of Education 1 3.5 Years of Education 2 3.6 Hours Worked 3.7 Capital Share 4 Quality of Human Capital 4.1 Quality of Schooling: Inputs 4.1.1 Teachers’ Human Capital 4.1.2 Pupil-Teacher Ratios 4.1.3 Spending 4.2 Quality of Schooling: Test Scores 4.3 Experience 4.4 Health 4.5 Social vs. Private Returns to Schooling and Health 5 Quality of Physical Capital 5.1 Composition 5.2 Vintage Effects 5.3 Further Problems with K 6 Sectorial Differences in TFP 6.1 Industry Studies 6.2 The Role of Agriculture 6.3 Sectorial Composition and Development Accounting 7 Non-Neutral Differences 7.1 Basic Concepts and Qualitative Results 7.2 Development Accounting with Non-Neutral Differences 8 Conclusions
8 Conclusions Development accounting is a powerful tool to getting started thinking about the sources of income differences across countries. As of now, the answer to the development-accounting question — do observed differences in the factors employed in production explain most of the cross-country variation in income — is: no, way no. This negative answer is robust to attempts to improve the measurement of human capital by allowing for differences in the quality of schooling and in health status of the population; to attempts to account for the age composition of the capital stock; to sectorial disaggregations of output; and to several other robustness checks. On the other hand, incomplete knowledge about certain key parameters that describe the relationship between inputs and outputs implies that the jury should be treated as being still out. For one thing, depending on the elasticity of substitution between capital of different 61 types, the observed wild heterogeneity in the composition of the capital stock by type of equipment could turn out to be a key proximate determinant of income differences. For another, depending on the elasticity of substitution between physical and human capital, we may find that all is needed for these factors to explain a large fraction of income inequality is a departure from Cobb-Douglas. Disaggregating the government sector out of the data may also potentially reduce the unexplained component of GDP. There is no deep reason why we should not be able to make progress on these three fronts, so that my assessment of the future of this research enterprise is optimistic.
Comments
Post a Comment